Where drones actually pay back in India — and why it isn't last-mile pizza.

Arjun opens the conversation with the most under-quoted line in Indian deep-tech right now. By his count there are roughly 250 registered drone outfits, but the vast majority will not earn revenue at scale because they have started from the airframe rather than the problem — designing a drone first and looking for a use-case second. The right sequence, he insists, is the inverse: read the industry, find the workflow where drones bring a measurable efficiency, then build the aircraft around that workflow. "We don't do drones for the sake of doing drones," he says, more than once.

The framing matters because the photogenic part of the Indian drone story — the e-commerce demo videos, the wedding photography, the pizza delivery teasers — is essentially the funded part. Mid-mile logistics, oil and gas inspections, agri spraying and defence are the unfunded parts, and they happen to be where the money lives. The asymmetry between cash flow and column inches is so severe that an observer relying only on press coverage would mis-rank the industry by a factor of ten.

The cleanest moment in the conversation. A retailer's hub sits just outside the city — for Bangalore, somewhere near Bommasandra or Devanahalli. Smaller fulfilment centres sit inside the city: ITPL Whitefield, Koramangala, JP Nagar. The stretch between those two points is the mid-mile, and it is the single piece of the supply chain where road geometry punishes the truck most. Arjun gives the specific number: 21 kilometres from Whitefield to the studio, one hour forty-five minutes by road. Same trip, point-to-point, takes fifteen minutes for an aircraft. Both endpoints are real estate the operator already controls, so there is no third-party doorstep problem and no last-mile consent overhead.

The reason this is a P&L and not a press release is that the route is repeatable. A truck waits to fill up; a drone runs flights every fifteen minutes regardless of utilisation. With two aircraft criss-crossing, the operator gets six flights an hour, no traffic risk, no fuel volatility, and an effective doubling of throughput on a corridor that the road network simply cannot improve. Last-mile is glamorous and unsolved; mid-mile is unglamorous and immediately economic.

Arjun is unsentimental about consumer drone delivery. The pizza-at-your-doorstep narrative does not survive contact with Indian density, with Indian airspace, or with Indian last-mile cost structure. He references the Amazon pilot in the United States where regulation requires a spotter every half mile; a twenty-mile route therefore needs forty observers, and the labour math destroys the unit economics. "If you're adding more humans to the drone, you might as well use a truck." The Western press talks last-mile because Western labour is expensive enough that drones offset it. India has no such offset.

The implication is that the entire Indian drone industry is going to spend a decade building competence in workflows that do not look like the foreign template. The future of Indian drone economics is depot-to-depot, hospital-to-lab, base-to-base — high-frequency point-to-point hops where both endpoints belong to the operator. Borrowing the Amazon Prime Air narrative is a categorical error.

From the early 2010s the Directorate General of Civil Aviation effectively grounded commercial drones — an eighteen-month moratorium followed by years of restrictive rules driven by the rise of DJI consumer airframes and irresponsible operators flying near defence airspace. Arjun supports the pause: it was the right call for a category that had become unorganised. The pivot came in stages. The 2018 framework was the first real opening. The Drone Rules 2021, issued in August of that year, abolished the old UIN regime, simplified type certification, and explicitly mapped green, yellow and red zones across the country — with most of India coloured green. The PLI scheme of ₹120 crore for drone manufacturers landed in September 2021. The BVLOS sandbox flights followed in 2022.

The investible reading: in regulated categories, the regulator's calendar leads the market's. Operators who timed their build to 2021 are now scaling; those who built ahead of it spent a decade burning capital, and those who waited until 2024 are now late. The rulebook is a directly readable demand signal.

The most useful operating note in the conversation. When Arjun first walks into a logistics customer, the question is never "how much does this drone cost." The right framing is fleet operation: how to manage fifty aircraft simultaneously, how to schedule charging cycles, how long the batteries last, how to load and unload safely, how to balance payload, what the rapid-response team looks like when a motor fails, where the alternate landing points sit. Selling one drone is a hobbyist transaction. Selling fifty is an infrastructure transaction. The first sale is the easy part; the deployment is the company.

The discipline this forces on the founding team is unusual. Scandron's pitch to a customer is structured the way an aircraft OEM pitches an airline: not "buy this aircraft" but "here is how you make money flying this aircraft." The drone manufacturer who can sit across from a logistics manager and walk through the operating cost of a fifty-unit fleet is, in 2026, almost the entire competitive field.

Arjun lands a quietly important point about the conversational vocabulary of the category. When a non-specialist asks about drones, the first question is almost always battery life. For a logistics drone that question is, in his words, irrelevant. The right metrics are payload and range — how heavy a parcel can the aircraft carry, and how far can it carry it. A 25kg cargo drone configured for short range carries 40 kg if the route is short. The same airframe configured for longer range drops to 5 kg payload at the limit. The trade-off is not battery vs. duration; it is payload vs. distance, modulated by altitude and ambient conditions.

The point generalises. Every emerging category has a default question the lay observer asks, and that question is almost always wrong. The serious operators are the ones who have built a different question and trained their customers to ask it.

One of the most quietly competitive observations of the conversation, and the cleanest articulation of why iDEX is structurally biased toward Indian drone firms. The Indian Army operates in theatres — Siachen, Ladakh, parts of the Northeast — where combat starts at 12,000 feet and climbs from there. Nobody else has this requirement at this scale. A Dutch firm selling a 200 kg cargo drone for eight minutes at sea level is, Arjun says, "designed to pick something up and put it on top of a building. They are not logistics drones." A high-altitude logistics aircraft — thin atmosphere, dual-altitude operation, 4,000m to 5,000m climb carrying 20 kg — is a globally unique product category.

The Make-in-India procurement angle becomes structural, not political. The Army needs a drone nobody else builds. Scandron, by spending the engineering hours on that variant, owns a buyer with a near-monopsony procurement appetite for at least a decade. The civilian platform is one product line; the defence variant is another, sharing IP but not chassis.

Arjun is dismissive of design fluff. A logistics drone, in his words, is "a truck of drones. A truck doesn't necessarily look pretty but it looks solid." Scandron's airframes are designed for sturdiness, easy field repair, modular swap-out of components, fast battery turnaround on the ground, and high uptime. The aesthetic choice is industrial — visible bolts, exposed wiring runs, no plastic skin pretending to be something else. The economic argument behind the aesthetic is that every aesthetic choice that adds time on the ground is a flight that did not happen, and the flight that did not happen is the revenue that did not arrive.

The deeper move is to refuse the consumer-grade frame. Most Indian drone companies started in the DJI-inspired consumer space and tried to scale up to industrial. Scandron started in industrial and refused the consumer-grade aesthetic from day one. The two design languages are fundamentally different products.

The oil and gas vertical produces the most concrete cost-saving story in the conversation. A petrochemical furnace or chimney — 100 to 120 metres high, the size of a small building — needs internal inspection during a shutdown. The legacy process takes about a month: drop temperature and pressure, flush with steam, flush with nitrogen, flush again, introduce oxygen, open the manhole, erect scaffolding, install lighting, send a human inspector. The plant is offline through the entire sequence at a cost Arjun puts at three to four crore rupees per day. Scandron's confined-space drone, carrying its own lighting system, completes the inspection in roughly half an hour from the moment the manhole opens. If there is no fault, the plant restarts in two days instead of a month.

The reason this is the cleanest sale in the Indian drone market is that the savings are denominated in crores per day, not lakhs per route. There is no need to sell the buyer on the future of logistics. The buyer is already paying for the alternative; the drone simply costs less.

Arjun's reading of the post-2021 airspace map is unsentimental. Except for specifically demarcated orange and red zones — airports, sensitive defence installations, certain national-security corridors — the entire country is a green zone in which drones can fly, with altitude caps and range caps that he considers sensible rather than restrictive. The 120-metre altitude ceiling mandated by DGCA is not a constraint; it is a coordination layer that makes traffic management tractable. The type-certification process exists because not every company building drones knows how to build a flightworthy one. When a truck fails it stops; when a drone fails it falls. The certification rigour is the price of operating at scale.

For a founder evaluating the regulatory environment, the green-zone map is the single most important document in the room. It tells you which corridors are open today, which become open with a permission application, and which require a separate compliance route. Most pitch decks do not contain that map. They should.

Beyond Visual Line of Sight — abbreviated BVLOS — is the single regulatory phrase that decides whether a commercial drone business is real or theoretical. Within visual line of sight, operations are recreational or short-range, and the unit economics never close. Beyond visual line of sight, the aircraft becomes infrastructure. Arjun explains the customer's first misunderstanding clearly: clients sometimes ask for a human pilot to remain in command throughout a long-range flight. He refuses. The aircraft is twenty kilometres away from the human; the human has no spatial perspective, no real-time orientation, no ability to react to wind shear that hasn't reached them yet. The right architecture is autonomous flight with a safety pilot monitoring telemetry, and failsafe behaviours baked into the aircraft — abort and return on wind exceedance, abort and return on battery anomaly, abort and return on communication loss.

The BVLOS sandbox, operational from 2022, is the slow road to scale. Permissions are requested per flight; the operator builds a track record; the regulator allows progressively wider corridors. By the end of this decade, Arjun's bet is that the sandbox becomes a routine permission, and that is when the mid-mile P&L compounds.

The honest underspoken concern in the industry. The DGCA mandates 120m altitude operation. With ten Indian drone companies flying commercial routes, that means traffic on a single altitude plane. Separation corridors, direction-of-travel conventions, vertical avoidance protocols — the entire Unified Traffic Management layer for drones is being built in real time, on the wing. Arjun describes Scandron's partnership with a UTM provider whose protocol allows any drone manufacturer to be detected in their system, as distinct from Honeywell and Thales whose stacks today are largely closed to their own equipment. He believes the open-protocol approach is what will scale.

The deeper observation: today's UTM is tracking, not avoidance. Active collision avoidance is built into individual aircraft but is not yet a coordinated network behaviour. Even commercial aircraft, he notes, don't all have avoidance — it is expensive to add. The thirty-second version is that the industry has built the first plane; it has not yet built the first air-traffic-control tower.

Arjun has been working in drones for seventeen years. The historical arc he walks through is useful because it explains why the current cohort of founders is not pioneering — it is the second wave. In 2006 there was no Indian drone policy because there were no civilian drones; the only customer was the defence establishment, and it operated under its own airspace umbrella. Operators called Air Traffic Control directly, NOTAMs were issued to airline pilots, drones flew. The aircraft themselves cost a hundred to a hundred and fifty thousand US dollars each because every component was custom-designed and sourced from outside India. The supplier base of motors, autopilots, sensors and chassis fabricators did not exist.

The two changes that broke the dam, in his telling, were the arrival of DJI's sub-two-thousand-dollar consumer drones and the parallel collapse in compute prices — particularly the Nvidia graphics chips that made photogrammetry economically tractable. Suddenly a single engineer with two lakh rupees could build a small drone and a GIS subscription and find a wedding photographer to pay them. That cohort created the irresponsible operators, who forced the regulatory clampdown, which in turn forced the 2018 framework, which in turn produced the 2021 reforms. The current liberal regime is the third or fourth iteration of an argument that started two decades ago.

The conversation is admirably honest about competitive context without naming every player. The visible Indian drone cohort — Garuda Aerospace, whose founder Agnishwar Jayaprakash received a Padma Shri in 2024 and whose work spans agri and UAV; ideaForge, listed on the NSE since September 2023 with a defence-dominant revenue mix; Marut Drones in medical and vaccine delivery; Aarav Unmanned Systems in mining surveying; AlphaTOCOL Engineering and Asteria Aerospace in defence-adjacent — is collectively producing the first generation of Indian drone IPOs and acquisitions. Arjun's positioning is deliberately orthogonal: not consumer, not surveying, not pure defence, but a mid-mile logistics service operator that owns the aircraft.

The implicit thesis is that of those eight or nine visible companies, only a handful will survive consolidation, and the ones that do will own structural moats — either deep defence procurement relationships, or topology-defined commercial corridors, or genuine manufacturing IP in components like motors and autopilots. The rest fold or get absorbed.

Arjun is realistic about Make-in-India for drones. Scandron designs the airframe and many of the electronics in-house; fasteners and chassis components are domestic; communication systems and autopilots are largely imported because foreign vendors have spent decades on dependability nobody local can yet match. ISRO has launched a homegrown autopilot, but new autopilots need hundreds of thousands of flight hours before the reliability curve stabilises. The same problem afflicts motors, batteries, and sensors. The path to indigenisation is to fly, modify, fly, modify — for years.

His specific test of policy maturity is the testing labs. There is, he notes, exactly one lab in India today large enough to certify Scandron's biggest airframe, and it is based in Bangalore. That bottleneck is invisible to the policy conversation but binding to the operating reality. Make-in-India is real, partial, and at least a decade from full.

Arjun draws the chemistry cleanly. Lithium polymer is the workhorse. Lithium ion is heavier but more energy-dense for the equivalent footprint. Solid-state is similar to LiPo but denser. Graphene is the frontier. Petrol — an internal-combustion engine — still beats all of them on raw energy density, which is why long-endurance military drones run engines. The catch is that petrol drones are louder, heavier, more dangerous over populated areas, and have higher failure modes. Electric is the only viable architecture for civilian mid-mile logistics over Indian cities, and the binding constraint of electric is energy per kilogram. Every kilogram of battery is a kilogram of payload not carried, and the curve is uncompromising.

The implication for the policy conversation: regulation is not the bottleneck Indian drones face. Energy density is. The day a meaningful improvement in cell chemistry arrives — whether solid-state at scale or a graphene breakthrough — the unit economics of every Indian drone company shift overnight. Until then, the operators are designing inside an envelope set by the physics, not the rulebook.

One of the more startling moments in the conversation, given the recent ideaForge listing. Arjun is clear: Scandron is largely self-funded and does not need to go to an IPO, because the only legitimate reasons to go public are an early-investor exit or genuine capital deployment, and neither applies. He observes that several Indian tech companies have IPO'd to subsidise loss-making operations — raising money simply to pay the bills — and calls that a non-sustainable business model. If you raise money, he argues, it should fund R&D, infrastructure, or new asset classes that contribute to revenue. The IPO route is one capital format among many, not the destination.

The framing is unusual because it inverts the standard founder narrative. Most Indian deep-tech founders describe an IPO as the validation moment. Arjun describes it as a question of utility: what would the money do that the operating cash flow won't do better. Until the answer is real, the public markets are a distraction.